BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Acoustic Factorisation for Robust Speech Recognition - Eric(Yongqi
 ang) Wang (University of Cambridge)
DTSTART:20130111T123000Z
DTEND:20130111T133000Z
UID:TALK41167@talks.cam.ac.uk
CONTACT:Catherine Breslin
DESCRIPTION:For many practical scenario\, speech recognition systems need 
 to be robust against multiple acoustic factors\, e.g.\, speaker and noise 
 differences. A conventional approach would be adapting acoustic models by 
 transforms estimated for each speaker and environment combination. However
 \, an ideal approach would be based on the concept of acoustic factorisati
 on\, where transforms are factorised such that each component transform on
 ly models one distinct acoustic factor. This gives flexibility for model a
 daptation\, e.g.\, rapid speaker adaptation in fast changing environments\
 , as demonstrated by  the experiments. There are a few options to construc
 t factorised transform: the component transforms can be constrained to hav
 e different forms such that  different distortions can be modelled separat
 ely\nand/or by imposing overlapping data such that the component transform
 s learn different attributes. These options as well as future works will b
 e discussed in the talk.
LOCATION:Department of Engineering - LR6
END:VEVENT
END:VCALENDAR